Aider Polyglot leaderboard
Aider Polyglot leaderboard: multi-language code-editing pass rates for each LLM, plotted against API price to reveal the best value coding models. Current leader: GPT-5 at 88 (% pass).
Cost vs quality
Each dot is a model. The dashed line is the value frontier — top-left is the best trade-off (high score, low cost).
Full leaderboard
OPENOPEN
OPEN
OPENOPEN
OPENOPENOPENOPENOPENOPEN
OPENOPENScores from Epoch AI, updated —. Cost is joined from our daily API pricing where a model matches; models we don't price still appear with their score. Always verify on the benchmark's own site before relying on a number.
What Aider Polyglot measures
The Aider Polyglot benchmark tests something coding assistants do all day: editing existing code and returning changes in a precise, machine-applicable format. It uses a set of the hardest Exercism programming exercises across multiple languages — including C++, Go, Java, JavaScript, Python and Rust — so it rewards models that are strong beyond just Python.
The score is the percentage of exercises solved, where a solution must both be correct and be delivered as a properly-formatted edit that applies cleanly. That format discipline matters: a model that writes good code but can’t reliably produce a valid diff will score lower — which mirrors real coding-assistant reliability.
How to read this page
Models are ranked by pass rate. As with SWE-bench, the cost-vs-quality chart uses each model’s output price per 1M tokens as the cost axis, since Aider’s public data doesn’t include a uniform per-task cost. The value frontier highlights models that combine a high pass rate with a low generation price.
Limitations
Scores are sensitive to the edit format a model is asked to use and can shift with Aider version changes. The tracked model set also tends to lag the very newest releases, so the freshest frontier models may not appear yet. Read it alongside LiveBench and SWE-bench rather than on its own.